Who Is Accountable When an AI Agent Writes Code?

Who is accountable when a GitHub Copilot agent raises a pull request? Learn how to review an AI agent PR and keep agents in your control. In this video (GH-600 Ep 4) you will learn: - Why the human, never the agent, is accountable for agent-written code - The 4 things a person must own: task, permissions, controls, merge approval - 4 anti-patterns that let an AI agent get out of control - The contributor model: no blind trust, no excessive suspicion - What to check in an agent PR: scope, goal, plan, rollback, success criteria, checks and artifacts - A live demo with CODEOWNERS, rulesets and merge approval Chapters: 0:00 Who Is Accountable for AI Agent Code? 0:42 4 Human Responsibilities Agents Never Own 3:07 Agent as a Ticket Assignee 3:36 4 Agent Anti-Patterns to Avoid 4:43 Contributor Model: Trust, but Verify 6:17 Plan, Scope & Rollback Traceability 7:00 Live Demo: Agent Issue and Plan 8:55 How to Review an Agent PR 10:23 CODEOWNERS and Merge Approval Demo 12:03 Key Takeaways GH-600 series so far: Ep 1 - 5 Differences Between AI Chatbots and Agentic Systems: https://www.youtube.com/watch?v=aP6BI48Yaac&list=PLGr9VtBrfNqg Ep 2 - Agentic AI Lifecycle | Plan to Pull Request: https://www.youtube.com/watch?v=wlrLMAemiOY Ep 3 - How to Control AI Agents in GitHub: https://www.youtube.com/watch?v=zaW9B6sS61U Watch next: Ep 3 (How to Control AI Agents in GitHub) covers the controls this video builds on. Full playlist: https://www.youtube.com/watch?v=aP6BI48Yaac&list=PLGr9VtBrfNqg 🎓 Courses: https://gauravkhurana.com/courses 🌐 Website: https://gauravkhurana.com 📅 1:1 mentoring: https://topmate.io/gauravkhurana #GitHubCopilot #AIAgents #GH600 #SoftwareTesting #SharingIsCaring